Research
No entries No research study is currently active.
Studies on language models and the systems around them, with their methods, results, and limitations.
Research
By Marco De Santis; work predating Nullsilver.
A study on whether reasoning models can learn from compressed versions of their own outputs, reducing the tokens they use without losing the steps needed to solve a problem.
From the entry
The evaluations are small, and shorter traces alone are not enough to show that a model reasons better.